Assignment of resource criticality scores to cloud resources based on cloud resource class

Abstract

In an aspect, a machine-learning (ML)-based classifier or regressor associated with a respective cloud resource class by is trained inputting information samples and resource criticality scores for the respective cloud resource class as training data. In a further aspect, the ML-based classifier or regressor is further utilized to assign a resource criticality score to a particular cloud resource in the respective cloud resource class.

What is claimed is: 
     
         1 . A method of operating a training component, comprising:
 receiving information samples that characterize attributes of a set of cloud resources of a cloud network,
 wherein the cloud network comprises a frontend platform and a backend platform, 
 wherein the frontend platform comprises frontend client infrastructure for interfacing with clients, 
 wherein the backend platform comprises backend platform infrastructure that comprises a group of distributed and interconnected computing devices with shareable hardware and/or software resources that support distributed implementation of a set of cloud applications via a respective set of cloud resources, 
 wherein each cloud resource of the set of cloud resources is associated with a set of shareable hardware and/or software resources of the backend platform, and 
 wherein each cloud resource of the set of cloud resources is associated with a cloud resource class that corresponds to one of a plurality of cloud resource classes supported by the cloud network; 
 assigning resource criticality scores for each cloud resource class associated with the set of cloud resource classes based on the information samples and rule-based heuristics; and 
   training, for each cloud resource class associated with the set of cloud resource classes, a machine-learning (ML)-based classifier or regressor associated with the respective cloud resource class by inputting the information samples and the resource criticality scores for the respective cloud resource class as training data.   
     
     
         2 . The method of  claim 1 , wherein the assigning comprises, for each cloud resource of the set of cloud resources:
 determining a probability distribution of a set of vectorized data attributes associated with the cloud resource associated with the cloud resource class;   assigning a score to each vectorized data attribute in the set of vectorized data attributes based on the probability distribution;   scaling each score in accordance with a cloud resource class-specific scaling factor; and   normalizing each scaled score for each cloud resource class to produce cloud resource class-specific resource criticality scores.   
     
     
         3 . The method of  claim 2 , wherein the probability distribution is associated with one or more lookup tables, one or more language tokens, or both. 
     
     
         4 . The method of  claim 2 , wherein the ML-based classifier or regressor is trained based on the probability distribution, the set of vectorized data attributes, and the cloud resource class-specific resource criticality scores. 
     
     
         5 . The method of  claim 2 , wherein the set of vectorized data attributes is obtained by:
 passing a first set of information samples in textual form through a ML-based sentence transformer to produce a set of fixed-length text vectors, or   transforming a second set of information samples into n-gram representations, or   a combination thereof.   
     
     
         6 . The method of  claim 2 , wherein the set of vectorized data attributes comprises vectorized data that is vectorized from the information samples. 
     
     
         7 . The method of  claim 6 ,
 wherein the vectorized data comprises data that is produced via a ML-based sentence transformer and the ML-based classifier or regressor comprises a neural network (NN), or   wherein the vectorized data comprises n-gram representations and the ML-based classifier or regressor is configured to perform logistic regression on the n-gram representations.   
     
     
         8 . The method of  claim 1 , further comprising:
 refining, for at least one cloud resource class associated with the set of cloud resource classes, the ML-based classifier or regressor associated with the at least one cloud resource class by inputting new information samples and new resource criticality scores for the at least one cloud resource class as new training data.   
     
     
         9 . The method of  claim 1 , wherein the information samples comprise:
 cost information, or   software information, or   purpose information, or   security control information, or   customer-defined tag information, or   component-specific configuration information, or   cloud network configuration information,   information extracted from an operating system (OS) of the component, or   any combination thereof.   
     
     
         10 . The method of  claim 9 , wherein the information samples are associated with a JavaScript Object Notation (JSON) configuration file. 
     
     
         11 . The method of  claim 1 , wherein at least one cloud resource cloud of the set of cloud resources corresponds to:
 an Azure virtual machine, or   an Amazon Web Services (AWS) instance, or   a Google Cloud Processing (GCP) instance, or   a relational database (RDB), or   a storage bucket or storage container, or   a billable resource type, or   a combination thereof.   
     
     
         12 . The method of  claim 1 , wherein the cloud network is a public cloud network, a private cloud network, a hybrid cloud network, or a multicloud network. 
     
     
         13 . A training component, comprising:
 one or more memories;   one or more transceivers; and   one or more processors communicatively coupled to the one or more memories and the one or more transceivers, the one or more processors, either alone or in combination, configured to:
 receive, via the one or more transceivers, information samples that characterize attributes of a set of cloud resources of a cloud network, 
 wherein the cloud network comprises a frontend platform and a backend platform, 
 wherein the frontend platform comprises frontend client infrastructure for interfacing with clients, 
 wherein the backend platform comprises backend platform infrastructure that comprises a group of distributed and interconnected computing devices with shareable hardware and/or software resources that support distributed implementation of a set of cloud applications via a respective set of cloud resources, 
 wherein each cloud resource of the set of cloud resources is associated with a set of shareable hardware and/or software resources of the backend platform, and 
 wherein each cloud resource of the set of cloud resources is associated with a cloud resource class that corresponds to one of a plurality of cloud resource classes supported by the cloud network; 
 assign resource criticality scores for each cloud resource class associated with the set of cloud resource classes based on the information samples and rule-based heuristics; and 
 train, for each cloud resource class associated with the set of cloud resource classes, a machine-learning (ML)-based classifier or regressor associated with the respective cloud resource class by inputting the information samples and the resource criticality scores for the respective cloud resource class as training data. 
   
     
     
         14 . The training component of  claim 13 , wherein the assigning comprises, for each cloud resource of the set of cloud resources:
 determine a probability distribution of a set of vectorized data attributes associated with the cloud resource associated with the cloud resource class;   assign a score to each vectorized data attribute in the set of vectorized data attributes based on the probability distribution;   scale each score in accordance with a cloud resource class-specific scaling factor; and   normalize each scaled score for each cloud resource class to produce cloud resource class-specific resource criticality scores.   
     
     
         15 . The training component of  claim 14 , wherein the probability distribution is associated with one or more lookup tables, one or more language tokens, or both. 
     
     
         16 . The training component of  claim 14 , wherein the ML-based classifier or regressor is trained based on the probability distribution, the set of vectorized data attributes, and the cloud resource class-specific resource criticality scores. 
     
     
         17 . The training component of  claim 14 , wherein the set of vectorized data attributes is obtained by:
 pass a first set of information samples in textual form through a ML-based sentence transformer to produce a set of fixed-length text vectors, or   transform a second set of information samples into n-gram representations, or   a combination thereof.   
     
     
         18 . The training component of  claim 14 , wherein the set of vectorized data attributes comprises vectorized data that is vectorized from the information samples. 
     
     
         19 . The training component of  claim 18 ,
 wherein the vectorized data comprises data that is produced via a ML-based sentence transformer and the ML-based classifier or regressor comprises a neural network (NN), or   wherein the vectorized data comprises n-gram representations and the ML-based classifier or regressor is configured to perform logistic regression on the n-gram representations.   
     
     
         20 . The training component of  claim 13 , wherein the one or more processors, either alone or in combination, are further configured to:
 refine, for at least one cloud resource class associated with the set of cloud resource classes, the ML-based classifier or regressor associated with the at least one cloud resource class by inputting new information samples and new resource criticality scores for the at least one cloud resource class as new training data.   
     
     
         21 . The training component of  claim 13 , wherein the information samples comprise:
 cost information, or   software information, or   purpose information, or   security control information, or   customer-defined tag information, or   component-specific configuration information, or   cloud network configuration information,   information extracted from an operating system (OS) of the component, or   any combination thereof.   
     
     
         22 . The training component of  claim 21 , wherein the information samples are associated with a JavaScript Object Notation (JSON) configuration file. 
     
     
         23 . The training component of  claim 13 , wherein at least one cloud resource cloud of the set of cloud resources corresponds to:
 an Azure virtual machine, or   an Amazon Web Services (AWS) instance, or   a Google Cloud Processing (GCP) instance, or   a relational database (RDB), or   a storage bucket or storage container, or   a billable resource type, or   a combination thereof.   
     
     
         24 . The training component of  claim 13 , wherein the cloud network is a public cloud network, a private cloud network, a hybrid cloud network, or a multicloud network.